SE-U-Net: Contextual Segmentation by Loosely Coupled Deep Networks for Medical Imaging Industry

Lin Yi Jiang, Cheng Ju Kuo, O. Tang-Hsuan, Min Hsiung Hung, Chao Chun Chen

研究成果: Conference contribution

摘要

We proposed a context segmentation method for medical images via two deep networks, aiming at providing segmentation contexts and achieving better segmentation quality. The context in this work means the object labels for segmentation. The key idea of our proposed scheme is to develop mechanisms to elegantly transform object detection labels into the segmentation network structure, so that two deep networks can collaboratively operate with loosely-coupled manner. For achieving this, the scalable data transformation mechanisms between two deep networks need to be invented, including representation of contexts obtained from the first deep network and context importion to the second one. The experimental results reveal that the proposed scheme indeed performs superior segmentation quality.

原文English
主出版物標題Intelligent Information and Database Systems - 13th Asian Conference, ACIIDS 2021, Proceedings
編輯Ngoc Thanh Nguyen, Suphamit Chittayasothorn, Dusit Niyato, Bogdan Trawiński
發行者Springer Science and Business Media Deutschland GmbH
頁面678-691
頁數14
ISBN(列印)9783030732790
DOIs
出版狀態Published - 2021
事件13th Asian Conference on Intelligent Information and Database Systems, ACIIDS 2021 - Phuket, Thailand
持續時間: 2021 四月 72021 四月 10

出版系列

名字Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
12672 LNAI
ISSN(列印)0302-9743
ISSN(電子)1611-3349

Conference

Conference13th Asian Conference on Intelligent Information and Database Systems, ACIIDS 2021
國家/地區Thailand
城市Phuket
期間21-04-0721-04-10

All Science Journal Classification (ASJC) codes

  • 理論電腦科學
  • 電腦科學(全部)

指紋

深入研究「SE-U-Net: Contextual Segmentation by Loosely Coupled Deep Networks for Medical Imaging Industry」主題。共同形成了獨特的指紋。

引用此